Multi-person heartbeat detection and positioning method and system based on simo radar

Abstract

The application discloses a multi-person heartbeat detection and positioning method and system based on a SIMO radar, and the method comprises the following steps: performing amplitude and phase correction on echo signals of each channel of the radar; performing weighting on the corrected signals to obtain echo signals corresponding to search angles; calculating two-dimensional FFT of the echo signals in the range-doppler domain, and performing reconstruction; imaging to obtain distance and angle information of human targets; calculating an optimal directional diagram weight vector through the DBF principle, and then calculating the range-Doppler FFT; obtaining vital sign signals of each human target; estimating the heart rate by using PCA and ZASLMS algorithms, and then extracting the vital sign signals according to the estimated heart rate by using a tensor decomposition theory algorithm, so as to obtain target heartbeat signals and restore corresponding time-domain waveforms. The method is effective and feasible, and has reliable performance, and can accurately determine the distance and angle information of multiple human targets located in different or same distance gates, suppresses the interference of moving and static objects, and extracts the heartbeat signals of corresponding targets.

Description

Technical Field

[0001] The present invention belongs to the field of radar technology, and in particular to a multi-person heartbeat detection and positioning method based on SIMO radar. Background Art

[0002] Existing research shows that Doppler radar is widely used in the field of detecting vital signs. This technology is also increasingly being used in medical treatment, emergency rescue, indoor monitoring, and other fields. Radar's ability to achieve long-term, stable detection has made it more attractive than video and infrared detection technologies. However, most current Doppler radars are only capable of stably monitoring the vital sign signals of a single target and are limited to accurately detecting breathing signals. With the continuous development of radar technology and increasing practical demand, the simultaneous positioning of multiple targets and detection of vital signs holds broad application prospects.

[0003] Currently, in the field of Doppler radar vital sign detection, many studies use principles such as MTI when locating target positions. These theories can only remove interference from stationary objects, but have not conducted in-depth research on removing interference from moving objects such as fans. Therefore, during positioning and imaging, they are easily interfered with by moving targets other than human targets, and moving objects are mistakenly judged as human targets to be detected. The document "Research on Radar Clutter Classification and Suppression Technology" also studied clutter interference and used the MTI principle to suppress background stationary clutter such as ground clutter and sea clutter. However, it did not study the interference suppression of moving interference objects on the detected target.

[0004] At the same time, there are multiple existing human vital signs signal processing algorithms, which focus on the research of respiratory signal waveforms and heart rate, and there is less research on the recovery of heart rate signal time domain waveforms. In the document "Research on Vital Sign Signal Detection Based on Millimeter Wave Radar", the research algorithm is limited to the detection of breathing and heart rate of two targets, and does not involve the research of heart rate signal waveforms; the document "Research on Life Signal Detection Technology Based on FMCW Millimeter Wave Radar" uses the empirical mode decomposition algorithm for the separation and extraction of heart rate signals, but the extraction results have mode aliasing problems and the signal-to-noise ratio is low. The final result analysis is also limited to frequency comparison, and does not involve comparison of time domain waveforms. Summary of the Invention

[0005] The object of the present invention is to provide a method for multi-person heartbeat detection and positioning based on SIMO radar to address the problems existing in the above-mentioned prior art.

[0006] The technical solution to achieve the purpose of the present invention is: a multi-person heartbeat detection and positioning method based on SIMO radar, the method comprising the following steps:

[0007] Step 1: calibrate each channel of the SIMO radar, and then calculate the steering vector of each search angle between -α and α to weight the calibrated signal to obtain the echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected;

[0008] Step 2: Calculate the range Doppler two-dimensional FFT based on the echo signals of each search angle in step 1, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, image and obtain the angle θ of each human target to be measured. p , distance gate information d p ;

[0009] Step 3: According to the target angle information θ obtained in step 2 p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital sign signal within the angle where the target is located;

[0010] Step 4: Get the vital sign signal according to step 3 and take the target angle θ p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the zero-attractive-sign minimum mean square error ZASLMS algorithm to estimate the target's heart rate f h ;

[0011] Step 5: Target heart rate f obtained in step 4 h ,Using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target.

[0012] Furthermore, the SIMO radar is calibrated for each channel as described in step 1. Then, for each angle between the search angle range -α and α, the steering vector of the corresponding angle is calculated to weight the calibrated signal and obtain the echo signal within the search angle range, specifically:

[0013] Step 1-1: Take the first receiving channel as the adjustment reference, calculate the amplitude ratio and phase difference of other channels relative to the reference, and obtain a set of amplitude and phase correction coefficients C adj :

[0014]

[0015] Where m=1,…,M, M is the number of receiving channels, b m is the amplitude value of the spectrum peak of the mth receiving channel, is the phase value of the spectrum peak of the mth receiving channel;

[0016] Calibrate each channel of SIMO radar:

[0017] B(k)=C adj A(k)

[0018] Where B(k) is the corrected signal of each channel, A(k) is the original signal of each channel, and A(k) = [s0(k), s1(k), ..., s m (k)],s m (k) is the sampling data of the mth channel, k is the sampling point of each channel;

[0019] Step 1-2: Calculate the corresponding steering vector a(θ) for each search angle θ within -α to α:

[0020] a(θ)=[1 e j2πdsin(θ) / λ … e j2π(m-1)dsin(θ) / λ ]

[0021] Where d is the array element spacing, θ is the search angle, m is the number of channels, and λ is the wavelength corresponding to the highest operating frequency;

[0022] Step 1-3: Using the steering vector a(θ) as the weighting vector, perform digital beamforming weighting on the corrected signal B(k) of each channel to obtain the echo signal s(k) of the search angle θ:

[0023] s(k)=a(θ) H B(k).

[0024] Furthermore, in step 2, the range Doppler two-dimensional FFT is calculated based on the echo signals of each search angle in step 1, and the interference of static objects and moving objects other than the human body is suppressed by using a two-dimensional sparse reconstruction algorithm. Finally, imaging is performed to obtain the angle θ of each human target to be measured. p , distance gate information d p , specifically including:

[0025] Step 2-1, calculate the FFT of the range-Doppler two-dimensional N points for the weighted echo signals s(k) of each search angle obtained in step 1, and obtain the matrix RTM[N,N];

[0026] Step 2-2, set a Doppler threshold f t , find the maximum value point in the matrix RTM[N,N], which is the target. If the Doppler of the corresponding target is higher than the Doppler threshold f t , it is considered to be an interfering moving target, and the radar echo signal at this angle is set to zero;

[0027] Step 2-3: If the target Doppler is lower than the set Doppler threshold f t First, calculate the IFFT of each row of the matrix RTM[N,N] and perform sparse reconstruction in the Doppler dimension with the sparsity set to 1. Then, sum each row to obtain a frequency domain column vector. Calculate the IFFT of this vector and perform sparse reconstruction in the range dimension with the sparsity also set to 1. This will obtain the one-dimensional frequency domain amplitude vector of the radar echo signal at this angle after suppressing the interference of stationary objects and multipath clutter.

[0028] Step 2-4, according to step 2-3, obtain the one-dimensional amplitude vector of the echo signal at each angle, arrange and combine each angle in sequence, and finally obtain the imaging diagram of the angle distance dimension. The corresponding imaging point is the location of the human target, and the angle θ of the human target is obtained. p With range gate information d p .

[0029] Furthermore, the sparse reconstruction algorithm steps in steps 2-3 are as follows:

[0030] Step 2-3-1, calculate IFFT for each row of RTM[N,N] to obtain N one-dimensional original sparse signal vectors f i , i=1,…,N;

[0031] Step 2-3-2, for f i Calculate the observations:

[0032] y=Ψf i

[0033] in, Gaussian random matrix is ​​used as the measurement matrix;

[0034] Step 2-3-3, calculate the perception matrix:

[0035]

[0036] Where Φ = (fft(eye(N,N))) -1 is the frequency domain sparse basis matrix, (·) -1 For the inverse operation, fft(·) is used to calculate the FFT of the matrix, and eye(N,N) is used to generate the N×N identity matrix;

[0037] Step 2-3-4, in this vital sign detection scenario, set the sparsity k = 1 and initialize the residual r0 = y;

[0038] Step 2-3-5, calculate the residual r and the columns in the perception matrix The inner product of , find the index λ of the maximum value i :

[0039]

[0040] Where r is the residual, is the column in the perception matrix, N is the number of FFT points;

[0041] Step 2-3-6, calculate the reconstruction set for this time Use the least squares method to calculate the original sparse signal f i The approximate solution of :

[0042]

[0043] Where, is the approximate solution of the signal, y is the observed value, Θ i To reconstruct the set, f i is the original sparse signal;

[0044] Step 2-3-7, obtain N approximate solutions After that, each vector is summed to form a one-dimensional distance column vector d θ (N×1);

[0045] Step 2-3-8, for d θ Calculate IFFT, and then perform steps 2-3-2 to 2-3-6 to obtain the current angle d θ The approximate solution of That is the one-dimensional amplitude vector after interference suppression at this angle.

[0046] Furthermore, in step 3, the angle information θ of the target obtained in step 2 is obtained. p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital signs signal within the target angle, including:

[0047] According to the calculation method of the steering vector in step 1-2, the search angle is limited to the angle θ where the pth human target is located. p , detected by the vital signs radar, at a distance of about 1.6m to 2.5m from the radar, the human body angle width is about ±8°, taking [θ p -8,θ p +8] angle range, with θ0 as the interval, each angle in the range corresponds to a steering vector, the echo signal B(k) is weighted, and the range dimension FFT is calculated, and then the range gate information d of the imaging target is obtained according to steps 2-4 p , extract the target's vital signs Total Angle components, n is the number of sampling points.

[0048] Furthermore, in step 4, the vital sign signal is obtained according to step 3, and the angle θ of the target is taken. p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the ZASLMS algorithm to estimate the target heart rate f h , specifically including:

[0049] Step 4-1, the vital sign signal LS[θ p ,n] performs low-pass filtering to remove high-frequency interference;

[0050] Step 4-2, the filtered signal LS1[θ p ,n]Use the segmented principal component analysis PCA method to extract the radar echo vital sign signal X2(n) containing the target's breathing and heartbeat;

[0051] Step 4-3, performing a differential operation on the vital sign signal X2(n);

[0052] y j =[X2(2)-X2(1) X2(3)-X2(2) … X2(n)-X2(n-1)]

[0053] Where n is the number of sampling points, X2(i) is the vital sign signal of the i-th sampling point, i = 1, 2, ..., n;

[0054] Step 4-4, take the differential signal y j As expected, the ZASLMS algorithm is used to iterate and obtain the heart rate f h .

[0055] Furthermore, the segmented principal component analysis (PCA) method used in step 4-2 specifically includes:

[0056] Segment the signal, select a certain window length, leave some margin before and after, and finally combine the signal to only need the window length part;

[0057] Step 4-2-1, take the filtered signal LS1[θ p ,n], the real part is the first row, the imaginary part is the second row, and the matrix X(2×n) of the vital sign data set is constructed. n is the number of signal sampling points, and X1 is obtained after preprocessing by removing the mean;

[0058] Step 4-2-2, find the covariance matrix of X1 Then perform eigenvalue decomposition on this covariance matrix:

[0059] R=UΛU T

[0060] Where Λ is a diagonal matrix composed of the eigenvalues ​​λ1 and λ2 of R, Λ = diag[λ1,λ2], and U is the corresponding eigenvector matrix;

[0061] Step 4-2-3, sort the obtained eigenvalues ​​from large to small, and take the eigenvector corresponding to the larger eigenvalue, that is, the 2×1 dimensional eigenvector u1;

[0062] Step 4-2-4, project the constructed vital sign data set onto the feature vector to obtain the required vital sign signal X2:

[0063] X 2(1×n) =u1 T X 1(2×n)

[0064] Where u1 is the eigenvector and X1 is the preprocessed signal matrix.

[0065] Furthermore, the differential signal y in step 4-4 is j As expected, the ZASLMS algorithm is used to iterate and obtain the heart rate f h , specifically including:

[0066] Step 4-4-1, calculate the basis matrix φ used by the ZASLMS algorithm j :

[0067]

[0068] Where M is the number of data sampling points, and N is the number of points for Fourier transform;

[0069] Step 4-4-2, initialize the expected output d(n) = y j , S(1)=0,y j is the vital sign differential signal calculated in step 4-3, m T (n) = φ j ;

[0070] Step 4-4-3, determine the vital sign differential signal y j After the expected output, each iteration needs to calculate the actual signal m T The error term of (n)S(n):

[0071] e(n)=d(n)-m T (n)S(n)

[0072] Where n is the number of sampling points, d(n) is the expected output, and m T (n) is the transformation basis matrix, S(n) is the final output, and e(n) is the error term;

[0073] Step 4-4-4, introduce the zero attraction sign factor -γsgn{S(n)}, and the final heartbeat spectrum iteration formula is:

[0074] S(n+1)=S(n)+μsgn{e(n)}m(n)-γsgn{S(n)}

[0075] Where n is the number of sampling points, S(n) is the final output, e(n) is the error term, m(n) is the transpose of the basis matrix, γ = μλ, λ and μ are the corresponding balancing factors, and sgn(·) is calculated as follows:

[0076]

[0077] Step 4-4-5, set the number of iterations c, and stop when the number of iterations reaches the set number. Finally, the spectrum of the heartbeat signal is obtained. The highest peak in the normal heart rate range of 1 to 2 Hz corresponds to the frequency f h This is your estimated target heart rate.

[0078] Furthermore, in step 5, the target heart rate f obtained in step 4 is h , using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target, specifically including:

[0079] Step 5-1, according to the above extraction Vital signs from different angles According to the principle of singular spectrum analysis, a trajectory matrix X is established for the signal at each angle. i , The sliding window length is L p , K p =nL p +1,

[0080]

[0081] x in the matrix j is the vital sign signal element, j = 1,…,n, and the sliding window length is generally the number of points containing the desired decomposition signal features;

[0082] Step 5-2, based on the trajectory matrix of each angle, combine them into a tensor

[0083] Step 5-3, perform tensor CP decomposition on the tensor to obtain r coefficients that can be regarded as eigenvalues, sort them from large to small, and each eigenvalue corresponds to a feature tensor

[0084] Step 5-4, using the matrix reconstruction idea in singular spectrum analysis, matrices Reconstruct information:

[0085]

[0086] Where s is the element in the matrix, N is the number of sampling points, L p is the sliding window length, K p =nL p +1;

[0087] Step 5-5, reconstruct the vector s n Sum up and obtain the corresponding vital sign signal reconstructed by the eigenvalue;

[0088] In step 5-6, the r decomposed components are screened based on the heart rate estimated in step 4, and the component with the closest frequency is selected as the target heartbeat signal waveform.

[0089] A multi-person heartbeat detection and positioning system based on SIMO radar, the system comprising:

[0090] The first module is configured to calibrate each channel of the SIMO radar, and then calculate a steering vector for each search angle within a search angle range of -α to α to weight the calibrated signal, thereby obtaining an echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected;

[0091] The second module is used to calculate the range Doppler two-dimensional FFT according to the echo signals of each search angle, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, imaging is performed to obtain the angle θ of each human target to be measured. p , distance gate information d p ;

[0092] The third module is used to obtain the target angle information θ according to the above p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital sign signal within the angle where the target is located;

[0093] The fourth module is used to obtain the target angle θ according to the vital sign signal obtained above. p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the zero-attractive-sign minimum mean square error ZASLMS algorithm to estimate the target's heart rate f h ;

[0094] The fifth module is used to obtain the target heart rate f h ,Using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target.

[0095] Compared with the existing technology, the present invention has the following significant advantages: 1) Through SIMO-FMCW radar, using digital beamforming technology, it can obtain the distance and angle information of multiple human targets within the detection range, and is applicable to a wider range of scenarios; 2) an improved two-dimensional sparse reconstruction compressed sensing algorithm is proposed to simultaneously suppress interference from stationary objects, multipath clutter interference, and moving objects, which can cleanly and accurately locate human targets within the imaging range; 3) a tensor decomposition algorithm based on singular spectrum analysis processing is proposed, which can accurately extract the heartbeat signal of the human target and more accurately restore the time domain waveform of the heartbeat, facilitating subsequent research.

[0096] The present invention is further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 This is a flow chart of the method for multi-person heartbeat detection and positioning based on SIMO radar of the present invention.

[0098] Figure 2 It is a segmented schematic diagram.

[0099] Figure 3 This is an imaging diagram using a traditional method in one embodiment, where -13° is an electric fan and 14° is a human target.

[0100] Figure 4 This is an improved sparse reconstructed imaging image in an embodiment where -13° is an electric fan and 14° is a human target.

[0101] Figure 5 This is an improved sparsely reconstructed imaging image of three targets in one embodiment.

[0102] Figure 6 Schematic diagram of a target heartbeat signal and a reference signal at -26° in an embodiment, FIG (a) is a time domain waveform comparison diagram, and FIG (b) is a signal spectrum comparison diagram.

[0103] Figure 7 Schematic diagram of a target heartbeat signal and a reference signal at -5° in an embodiment, FIG (a) is a time domain waveform comparison diagram, and FIG (b) is a signal spectrum comparison diagram.

[0104] Figure 8 Schematic diagram of a target heartbeat signal and a reference signal at 14° in an embodiment, FIG (a) is a time domain waveform comparison diagram, and FIG (b) is a signal spectrum comparison diagram. DETAILED DESCRIPTION

[0105] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0106] In addition, if the present invention has descriptions involving "first", "second", etc., the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0107] Combine Figure 1 , provides a method for multi-person heartbeat detection and positioning based on SIMO radar, the method comprising the following steps:

[0108] Step 1: calibrate each channel of the SIMO radar, and then calculate the steering vector of the corresponding angle for each search angle between -α and α to weight the calibrated signal and obtain the echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected; specifically, the following steps are performed:

[0109] Step 1-1: Take the first receiving channel as the adjustment reference, calculate the amplitude ratio and phase difference of other channels relative to the reference, and obtain a set of amplitude and phase correction coefficients C adj :

[0110]

[0111] Where m=1,…,M, M is the number of receiving channels, b m is the amplitude value of the spectrum peak of the mth receiving channel, is the phase value of the spectrum peak of the mth receiving channel;

[0112] Calibrate each channel of SIMO radar:

[0113] B(k)=C adj A(k)

[0114] Where B(k) is the corrected signal of each channel, A(k) is the original signal of each channel, and A(k) = [s0(k), s1(k), ..., s m (k)],s m(k) is the sampling data of the mth channel, k is the sampling point of each channel;

[0115] Step 1-2: Calculate the corresponding steering vector a(θ) for each search angle θ within -α to α:

[0116] a(θ)=[1 e j2πdsin(θ) / λ … e j2π(m-1)dsin(θ) / λ ]

[0117] Where d is the array element spacing, θ is the search angle, m is the number of channels, and λ is the wavelength corresponding to the highest operating frequency;

[0118] Step 1-3: Using the steering vector a(θ) as the weighting vector, perform digital beamforming weighting on the corrected signal B(k) of each channel to obtain the echo signal s(k) of the search angle θ:

[0119] s(k)=a(θ) H B(k)

[0120] Where B(k) is the corrected signal of each channel, θ is the search angle, and a(θ) is the weighted vector of angle θ.

[0121] Step 2: Calculate the range Doppler two-dimensional FFT based on the echo signals of each search angle in step 1, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, image and obtain the angle θ of each human target to be measured. p , distance gate information d p The specific process includes:

[0122] Step 2-1, calculate the FFT of the range-Doppler two-dimensional N points for the echo signals s(k) at each angle obtained by weighting in step 1, and obtain the matrix RTM[N,N];

[0123] Step 2-2, set a Doppler threshold f t (Based on experimental experience, the threshold should be greater than the Doppler f generated by normal human breathing. r , usually f t =2f r ), find the maximum value point in the matrix RTM[N,N], which is the target. If the Doppler of the corresponding target is higher than the Doppler threshold f t , it is considered to be an interfering moving target, and the radar echo signal at this angle is set to zero;

[0124] Step 2-3: If the target Doppler is lower than the set Doppler threshold f tFirst, calculate the IFFT of each row of the matrix RTM[N,N] and perform Doppler dimension sparse reconstruction with the sparsity set to 1. Then, sum each row to obtain a frequency domain column vector. Calculate the IFFT of this vector and perform sparse reconstruction in the range dimension with the sparsity also set to 1. This will obtain the one-dimensional frequency domain amplitude vector of the radar echo signal after interference suppression at this angle (because the sparsity is 1, after the signal is reconstructed, the amplitude at the point where the human target is located is not 0, while the amplitude at other points is 0, which also suppresses the interference of stationary objects and multipath clutter).

[0125] The specific steps of the sparse reconstruction algorithm are as follows:

[0126] Step 2-3-1, calculate IFFT for each row of RTM[N,N] to obtain N one-dimensional original sparse signal vectors f i (i=1,…,N);

[0127] Step 2-3-2, for f i Calculate the observed value y = Ψf i , Gaussian random matrix is ​​used as the measurement matrix;

[0128] Step 2-3-3, calculate the perception matrix Φ=(fft(eye(N,N))) -1 is the frequency domain sparse basis matrix, (·) -1 For the inverse operation, fft(·) is used to calculate the FFT of the matrix, and eye(N,N) is used to generate the N×N identity matrix;

[0129] Step 2-3-4, in this vital sign detection scenario, set the sparsity k = 1 and initialize the residual r0 = y;

[0130] Step 2-3-5, calculate the residual r and the columns in the perception matrix The inner product of , find the index λ of the maximum value i :

[0131]

[0132] Where r is the residual, is the column in the perception matrix, N is the number of FFT points;

[0133] Step 2-3-6, calculate the reconstruction set for this time The original sparse signal f can be calculated using the least squares method i The approximate solution of :

[0134]

[0135] Where, is the approximate solution of the signal, y is the observed value, Θ iTo reconstruct the set, f i is the original sparse signal.

[0136] Step 2-3-7, obtain N approximate solutions After that, each vector is summed to form a one-dimensional distance column vector d θ (N×1);

[0137] Step 2-3-8, for d θ Calculate IFFT, and then perform steps 2-3-2 to 2-3-6 to obtain the current angle d θ The approximate solution of That is the one-dimensional energy vector after interference suppression at this angle;

[0138] Step 2-4, according to step 2-3, obtain the one-dimensional amplitude vector of the echo signal at each angle, arrange and combine each angle in sequence, and finally obtain the imaging diagram of the angle distance dimension. The corresponding imaging point is the location of the human target, and the angle θ of the human target is obtained. p With range gate information d p .

[0139] Step 3: Based on the target angle information obtained in step 2, calculate the weight vector of the beam pointing to the angle to weight the echo signal, obtain the echo information of the target angle, calculate the distance dimension FFT, and obtain the vital signs signal within the target angle based on the target range gate information.

[0140] According to the calculation method of the steering vector in step 1-2, the search angle is limited to the angle θ where the pth human target is located. p , detected by the vital signs radar, at a distance of about 1.6m to 2.5m from the radar, the human body angle width is about ±8°, taking [θ p -8,θ p +8] angle range, with θ0 as the interval (0.1° in this experiment), each angle in the range corresponds to a steering vector, the echo signal B(k) is weighted, and the range dimension FFT is calculated, and then the range gate information d of the image target is obtained according to steps 2-4. p , the target's vital signs can be extracted Total Angle components, n is the number of sampling points.

[0141] Step 4: Get vital signs signal according to step 3 Take the target's angle vital sign signal LS[θ p ,n], segmented PCA is used to extract the target phase information, and then the target heart rate is estimated using the ZASLMS algorithm;

[0142] Step 4-1, the vital sign signal LS[θp ,n] first perform low-pass filtering to remove high-frequency interference;

[0143] Step 4-2, the filtered signal LS1[θ p ,n]Use segmented principal component analysis (PCA) to extract the radar echo vital sign signal X2(n) containing the target's breathing and heartbeat;

[0144] The segmentation principle diagram is as follows Figure 2 As shown;

[0145] Segment the signal, select a certain window length (windowlen), leave some overlap before and after, and finally combine the signal to only need the window length part;

[0146] The PCA algorithm calculation is as follows:

[0147] Step 4-2-1, take the filtered signal LS1[θ p ,n], the real part is the first row, the imaginary part is the second row, and the matrix X(2×n) of the vital sign data set is constructed. n is the number of signal sampling points, and X1 is obtained after preprocessing by removing the mean;

[0148] Step 4-2-2, find the covariance matrix of X1 Then perform eigenvalue decomposition on this covariance matrix:

[0149] R=UΛU T

[0150] Where Λ is a diagonal matrix composed of the eigenvalues ​​of R, Λ = diag[λ1,λ2], and U is the corresponding eigenvector matrix;

[0151] Step 4-2-3, sort the obtained eigenvalues ​​from large to small, and take the eigenvector corresponding to the larger eigenvalue, that is, the 2×1 dimensional eigenvector u1;

[0152] In step 4-2-4, project the constructed vital signs dataset onto the feature vector to obtain the required vital signs signal X2:

[0153] X 2(1×n) =u1 T X 1(2×n)

[0154] Where u1 is the eigenvector, X1 is the preprocessed signal matrix;

[0155] Step 4-3, performing a differential operation on the vital sign signal X2(n);

[0156] y j=[X2(2)-X2(1) X2(3)-X2(2) … X2(n)-X2(n-1)]

[0157] Where n is the number of sampling points, X2(n) is the sampled vital sign signal;

[0158] Step 4-4, take the differential signal y j For expectation, the zero attraction signed minimum mean square error (ZASLMS) algorithm is iterated to obtain the heart rate;

[0159] Step 4-4-1, calculate the basis matrix φ j :

[0160]

[0161] Where M is the number of data sampling points, and N is the number of points for Fourier transform;

[0162] Step 4-4-2, initialize the expected output d(n) = y j , S(1)=0,y j is the vital sign differential signal calculated in step 4-3, m T (n) = φ j ;

[0163] Step 4-4-3, determine the vital sign differential signal y j After the expected output, each iteration needs to calculate the actual signal m T The error term of (n)S(n):

[0164] e(n)=d(n)-m T (n)S(n)

[0165] Where n is the number of sampling points, d(n) is the expected output, and m T (n) is the transformation basis matrix, S(n) is the final output, and e(n) is the error term;

[0166] In step 4-4-4, in order to iterate the target heartbeat spectrum more quickly and accurately, compared with the traditional LMS algorithm, the zero attraction sign factor -γsgn{S(n)} is introduced. The final heartbeat spectrum iteration formula is:

[0167] S(n+1)=S(n)+μsgn{e(n)}m(n)-γsgn{S(n)}

[0168] Where n is the number of sampling points, S(n) is the final output, e(n) is the error term, m(n) is the transpose of the basis matrix, γ = μλ, λ and μ are the corresponding balancing factors, and sgn(·) is calculated as follows:

[0169]

[0170] In step 4-4-5, set the number of iterations, c, and stop when the number of iterations reaches the set number. Finally, the spectrum of the heartbeat signal is obtained. The frequency corresponding to the highest peak in the normal heart rate range of 1-2Hz is the estimated target heart rate.

[0171] Step 5: Based on the target heart rate obtained in step 4, the echo vital sign signal is processed using the tensor decomposition theory algorithm. Decompose and extract the heartbeat signal waveform of the human target.

[0172] Step 5-1, according to the above extraction Vital signs from different angles According to the principle of singular spectrum analysis, a trajectory matrix X is established for the signal at each angle. i , The sliding window length is L p , K p =nL p +1,

[0173]

[0174] x in the matrix i (i=1,…,n) is the vital sign signal element. The sliding window length is generally taken as the number of points containing the desired decomposition signal features. In this experiment, the heartbeat signal is decomposed with a sampling frequency of 200Hz, so two cycles of the heartbeat signal (L p =400);

[0175] Step 5-2: After obtaining the trajectory matrix of each angle, combine it into a tensor

[0176] Step 5-3, perform tensor CP decomposition on the tensor to obtain r coefficients that can be regarded as eigenvalues, sorting them from large to small, and each eigenvalue corresponds to a feature tensor

[0177] Step 5-4, using the matrix reconstruction idea in singular spectrum analysis, matrices Reconstruct information:

[0178]

[0179] Where s is the element in the matrix, N is the number of sampling points, L p is the sliding window length, K p =nL p +1;

[0180] Step 5-5, reconstruct the vector s n Sum up and obtain the corresponding vital sign signal reconstructed by the eigenvalue;

[0181] In step 5-6, the r decomposed components are screened based on the heart rate estimated in step 4, and the component with the closest frequency is selected as the target heartbeat signal waveform.

[0182] A multi-person heartbeat detection and positioning system based on SIMO radar, the system comprising:

[0183] The first module is configured to calibrate each channel of the SIMO radar, and then calculate a steering vector for each search angle within a search angle range of -α to α to weight the calibrated signal, thereby obtaining an echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected;

[0184] The second module is used to calculate the range Doppler two-dimensional FFT according to the echo signals of each search angle, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, imaging is performed to obtain the angle θ of each human target to be measured. p , distance gate information d p ;

[0185] The third module is used to obtain the target angle information θ according to the above p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital sign signal within the angle where the target is located;

[0186] The fourth module is used to obtain the target angle θ according to the vital sign signal obtained above. p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the zero-attractive-sign minimum mean square error ZASLMS algorithm to estimate the target's heart rate f h ;

[0187] The fifth module is used to obtain the target heart rate f h ,Using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target.

[0188] For the specific definition of the multi-person heartbeat detection and positioning system based on SIMO radar, please refer to the definition of the multi-person heartbeat detection and positioning method based on SIMO radar above, which will not be repeated here. Each module in the above-mentioned multi-person heartbeat detection and positioning system based on SIMO radar can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0189] The present invention uses a single-transmitter, multiple-receiver FMCW radar to achieve positioning of multiple targets, heartbeat signal detection, and waveform recovery while suppressing clutter and multipath interference from moving and static objects. It has high accuracy and is applicable to a wide range of scenarios.

[0190] The present invention will be further described in detail below with reference to the embodiments.

[0191] Example

[0192] Combine Figure 1 The present invention provides a method for detecting and locating multiple heartbeats based on SIMO radar, including the following contents:

[0193] 1. In this embodiment, a single-transmitter, eight-receiver FMCW radar is used. The radar carrier frequency is 5.8 GHz, the bandwidth is 400 MHz, and the sampling frequency is 200 kHz. A signal generator is connected to the transmitting antenna and placed 2 meters in front of the radar. A single-frequency continuous wave signal with a frequency of 5.8 GHz + 5 kHz is transmitted. Each receiving channel of the single-transmitter, multiple-receiver FMCW radar receives the echo signal, and the amplitude and phase correction coefficients are calculated to eliminate the inherent error of each receiving channel of the single-transmitter, multiple-receiver FMCW radar. After calibrating each channel of the single-transmitter, multiple-receiver FMCW radar, the steering vector of the corresponding angle is calculated at each angle between -60° and 60° to weight the corrected signal and obtain the echo signal within the search angle;

[0194] 2. Calculate the range Doppler two-dimensional FFT of the echo signal at each search angle, use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body, and finally image to obtain the angle and distance information of each human target to be measured; Figure 3 As shown in the figure, at -13° there is an electric fan, which is considered as a moving target interference, and at 14° there is a human target to be measured. The traditional imaging method cannot remove the non-human moving target interference. After imaging by the two-dimensional sparse reconstruction algorithm, the clutter interference and moving target interference are removed, as shown in the figure. Figure 4 shown. Figure 5 Shown are the two-dimensional sparse reconstruction algorithm images of three human targets to be tested.

[0195] 3. Based on the target angle information obtained in step 2, calculate the weight vector of the beam pointing to that angle to weight the echo signal, obtain the echo information at the target angle, calculate the range dimension FFT, and obtain the target's vital sign signal based on the target range gate information;

[0196] 4. Obtain the target's echo vital sign signal according to step 3, use segmented PCA to extract the target's phase information, and then use the ZASLMS algorithm to estimate the target's heart rate;

[0197] 5. Based on the target heart rate obtained in step 4, the tensor decomposition theory algorithm is used to decompose the echo signal to extract the heartbeat signal and waveform of the human target. Figure 6 The heartbeat waveform and heart rate of the human target at -26° and the comparison with the reference signal, Figure 7 The heartbeat waveform and heart rate of the human target at -5° and the comparison with the reference signal, Figure 8 The heartbeat waveform and heart rate of a human target at 14° are compared with the reference signal.

[0198] From the above, it can be seen that the method of the present invention is effective and feasible, with reliable performance. It can accurately determine the distance and angle information of multiple human targets located in different range gates or the same range gate, suppress various unnecessary clutter interferences and interference from moving targets other than human targets, and extract the corresponding heartbeat signal and waveform.

[0199] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-person heartbeat detection and positioning method based on SIMO radar, characterized in that: The method comprises the following steps: Step 1: calibrate each channel of the SIMO radar, and then calculate the steering vector of each search angle between -α and α to weight the calibrated signal to obtain the echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected; Step 2: Calculate the range Doppler two-dimensional FFT based on the echo signals of each search angle in step 1, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, image and obtain the angle θ of each human target to be measured. p , distance gate information d p ; Step 3: According to the target angle information θ obtained in step 2 p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital sign signal within the angle where the target is located; Step 4: Get the vital sign signal according to step 3 and take the target angle θ p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the zero-attractive-sign minimum mean square error ZASLMS algorithm to estimate the target's heart rate f h ; Step 5: Target heart rate f obtained in step 4 h ,Using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target.

2. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 1 is characterized in that: As described in step 1, the SIMO radar is calibrated for each channel. Then, for each angle in the search angle range -α to α, the steering vector of the corresponding angle is calculated to weight the calibrated signal and obtain the echo signal within the search angle range. Specifically, Step 1-1: Take the first receiving channel as the adjustment reference, calculate the amplitude ratio and phase difference of other channels relative to the reference, and obtain a set of amplitude and phase correction coefficients C adj : Where m=1,…,M, M is the number of receiving channels, b m is the amplitude value of the spectrum peak of the mth receiving channel, is the phase value of the spectrum peak of the mth receiving channel; Calibrate each channel of SIMO radar: B(k)=C adj A(k) Where B(k) is the corrected signal of each channel, A(k) is the original signal of each channel, and A(k) = [s0(k), s1(k), ..., s m (k)],s m (k) is the sampling data of the mth channel, k is the sampling point of each channel; Step 1-2: Calculate the corresponding steering vector a(θ) for each search angle θ within -α to α: a(θ)=[1 e j2πdsin(θ) / λ … And j2π(m-1)dsin(θ) / λ ] Where d is the array element spacing, θ is the search angle, m is the number of channels, and λ is the wavelength corresponding to the highest operating frequency; Step 1-3: Using the steering vector a(θ) as the weighting vector, perform digital beamforming weighting on the corrected signal B(k) of each channel to obtain the echo signal s(k) of the search angle θ: s(k)h(θ) H B(k)。 3. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 2 is characterized in that: Step 2 calculates the range Doppler two-dimensional FFT based on the echo signals of each search angle in step 1, and uses a two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects other than the human body, and finally images to obtain the angle θ of each human target to be measured. p , distance gate information d p , specifically including: Step 2-1, calculate the FFT of the range-Doppler two-dimensional N points for the weighted echo signals s(k) of each search angle obtained in step 1, and obtain the matrix RTM[N,N]; Step 2-2, set a Doppler threshold f t , find the maximum value point in the matrix RTM[N,N], which is the target. If the Doppler of the corresponding target is higher than the Doppler threshold f t , it is considered to be an interfering moving target, and the radar echo signal at this angle is set to zero; Step 2-3: If the target Doppler is lower than the set Doppler threshold f t First, calculate the IFFT of each row of the matrix RTM[N,N] and perform sparse reconstruction in the Doppler dimension with the sparsity set to 1. Then, sum each row to obtain a frequency domain column vector. Calculate the IFFT of this vector and perform sparse reconstruction in the range dimension with the sparsity also set to 1. This will obtain the one-dimensional frequency domain amplitude vector of the radar echo signal at this angle after suppressing the interference of stationary objects and multipath clutter. Step 2-4, according to step 2-3, obtain the one-dimensional amplitude vector of the echo signal at each angle, arrange and combine each angle in sequence, and finally obtain the imaging diagram of the angle distance dimension. The corresponding imaging point is the location of the human target, and the angle θ of the human target is obtained. p With range gate information d p .

4. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 3 is characterized in that: The steps of the sparse reconstruction algorithm in steps 2-3 are as follows: Step 2-3-1, calculate IFFT for each row of RTM[N,N] to obtain N one-dimensional original sparse signal vectors f i , i=1,…,N; Step 2-3-2, for f i Calculate the observations: y=Ψf i in, Gaussian random matrix is ​​used as the measurement matrix; Step 2-3-3, calculate the perception matrix: Where Φ = (fft(eye(N,N))) -1 is the frequency domain sparse basis matrix, (·) -1 For the inverse operation, fft(·) is used to calculate the FFT of the matrix, and eye(N,N) is used to generate the N×N identity matrix; Step 2-3-4, in this vital sign detection scenario, set the sparsity k = 1 and initialize the residual r0 = y; Step 2-3-5, calculate the residual r and the columns in the perception matrix Inner product, find the index λ of the maximum value i : Where r is the residual, is the column in the perception matrix, N is the number of FFT points; Step 2-3-6, calculate the reconstruction set for this time Use the least squares method to calculate the original sparse signal f i The approximate solution of : Where, is the approximate solution of the signal, y is the observed value, Θ i To reconstruct the set, f i is the original sparse signal; Step 2-3-7, obtain N approximate solutions After that, each vector is summed to form a one-dimensional distance column vector d θ (N×1); Step 2-3-8, calculate IFFT for dθ, and then perform steps 2-3-2 to 2-3-6 to obtain the approximate solution of dθ at the current angle. That is the one-dimensional amplitude vector after interference suppression at this angle.

5. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 4 is characterized in that: Step 3: The target angle information θ obtained in step 2 p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital signs signal within the target angle, including: According to the calculation method of the steering vector in step 1-2, the search angle is limited to the angle θ where the pth human target is located. p , detected by the vital signs radar, at a distance of 1.6m to 2.5m from the radar, the human body angle width is ±8°, and [θ p -8,θ p +8] angle range, with θ0 as the interval, each angle in the range corresponds to a steering vector, the echo signal B(k) is weighted, and the range dimension FFT is calculated, and then the range gate information d of the imaging target is obtained according to steps 2-4 p , extract the target's vital signs Total Angle components, n is the number of sampling points.

6. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 5 is characterized in that: Step 4: Get the vital sign signal according to step 3, and take the target angle θ p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the ZASLMS algorithm to estimate the target heart rate f h , specifically including: Step 4-1, the vital sign signal LS[θ p ,n] performs low-pass filtering to remove high-frequency interference; Step 4-2, the filtered signal LS1[θ p ,n]Use the segmented principal component analysis PCA method to extract the radar echo vital sign signal X2(n) containing the target's breathing and heartbeat; Step 4-3, performing a differential operation on the vital sign signal X2(n); y j =[X2(2)-X2(1)X2(3)-X2(2)…X2(n)-X2(n-1)] Where n is the number of sampling points, X2(i) is the vital sign signal of the i-th sampling point, i = 1, 2, ..., n; Step 4-4, take the differential signal y j As expected, the ZASLMS algorithm is used to iterate and obtain the heart rate f h .

7. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 6 is characterized in that: The segmented principal component analysis (PCA) method used in step 4-2 specifically includes: Segment the signal, select a certain window length, leave some margin before and after, and finally combine the signal to only need the window length part; Step 4-2-1, take the filtered signal LS1[θ p ,n], the real part is the first row, the imaginary part is the second row, and the matrix X(2×n) of the vital sign data set is constructed. n is the number of signal sampling points, and X1 is obtained after preprocessing by removing the mean; Step 4-2-2, find the covariance matrix of X1 Then perform eigenvalue decomposition on this covariance matrix: R=UΛU T Where Λ is a diagonal matrix composed of the eigenvalues ​​λ1 and λ2 of R, Λ = diag[λ1,λ2], and U is the corresponding eigenvector matrix; Step 4-2-3, sort the obtained eigenvalues ​​from large to small, and take the eigenvector corresponding to the larger eigenvalue, that is, the 2×1 dimensional eigenvector u1; Step 4-2-4, project the constructed vital sign data set onto the feature vector to obtain the required vital sign signal X2: X 2(1×n) =u1 T X 1(2×n) Where u1 is the eigenvector and X1 is the preprocessed signal matrix.

8. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 7 is characterized in that: The differential signal y j As expected, the ZASLMS algorithm is used to iterate and obtain the heart rate f h , specifically including: Step 4-4-1, calculate the basis matrix φ used by the ZASLMS algorithm j : Where M is the number of data sampling points, and N is the number of points for Fourier transform; Step 4-4-2, initialize the expected output d(n) = y j , S(1)=0,y j is the vital sign differential signal calculated in step 4-3, m T (n) = φ j ; Step 4-4-3, determine the vital sign differential signal y j After the expected output, each iteration needs to calculate the actual signal m T The error term of (n)S(n): e(n)=d(n)-m T (n)S(n) Where n is the number of sampling points, d(n) is the expected output, and m T (n) is the transformation basis matrix, S(n) is the final output, and e(n) is the error term; Step 4-4-4, introduce the zero attraction sign factor -γsgn{S(n)}, and the final heartbeat spectrum iteration formula is: S(n+1)=S(n)+μsgn{e(n)}m(n)-γsgn{S(n)} Where n is the number of sampling points, S(n) is the final output, e(n) is the error term, m(n) is the transpose of the basis matrix, γ = μλ, λ and μ are the corresponding balancing factors, and sgn(·) is calculated as follows: Step 4-4-5, set the number of iterations c, and stop when the number of iterations reaches the set number. Finally, the spectrum of the heartbeat signal is obtained. The highest peak in the normal heart rate range of 1 to 2 Hz corresponds to the frequency f h This is your estimated target heart rate.

9. The multi-person heartbeat detection and positioning method based on SIMO radar according to claim 8, characterized in that: Step 5: The target heart rate f obtained in step 4 h , using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target, specifically including: Step 5-1, according to the above extraction Vital signs from different angles According to the principle of singular spectrum analysis, a trajectory matrix X is established for the signal at each angle. i , The sliding window length is L p , K p =nL p +1, x in the matrix j is the vital sign signal element, j = 1,…,n, and the sliding window length is the number of points containing the desired decomposition signal features; Step 5-2, based on the trajectory matrix of each angle, combine them into a tensor Step 5-3, perform tensor CP decomposition on the tensor to obtain r coefficients that can be regarded as eigenvalues, sort them from large to small, and each eigenvalue corresponds to a feature tensor Step 5-4, using the matrix reconstruction idea in singular spectrum analysis, matrices Reconstruct information: Where s is the element in the matrix, N is the number of sampling points, L p is the sliding window length, K p =nL p +1; Step 5-5, reconstruct the vector s n Sum up and obtain the corresponding vital sign signal reconstructed by the eigenvalue; In step 5-6, the r decomposed components are screened based on the heart rate estimated in step 4, and the component with the closest frequency is selected as the target heartbeat signal waveform.

10. A multi-person heartbeat detection and positioning system based on SIMO radar, characterized in that: The system comprises: The first module is configured to calibrate each channel of the SIMO radar, and then calculate a steering vector for each search angle within a search angle range of -α to α to weight the calibrated signal, thereby obtaining an echo signal within the search angle range; the search angle range includes the detection angle range of each human target to be detected; The second module is used to calculate the range Doppler two-dimensional FFT according to the echo signals of each search angle, and use the two-dimensional sparse reconstruction algorithm to suppress the interference of static and moving objects except the human body. Finally, imaging is performed to obtain the angle θ of each human target to be measured. p , distance gate information d p ; The third module is used to obtain the target angle information θ according to the above p , calculate the weight vector a(θ) of the beam pointing to the angle to weight the echo signal, obtain the echo information s(k) at the angle where the target is located, calculate the range dimension FFT, and use the target range gate information d p , obtain the vital sign signal within the angle where the target is located; The fourth module is used to obtain the target angle θ according to the vital sign signal obtained above. p The vital sign signal LS[θ p ,n], using segmented PCA to extract the target phase information, and then using the zero-attractive-sign minimum mean square error ZASLMS algorithm to estimate the target's heart rate f h ; The fifth module is used to obtain the target heart rate f h ,Using the tensor decomposition theory algorithm, the echo vital sign signal is decomposed to extract the heartbeat signal waveform of the human target.

Images